Grading Carotid Stenosis With Ultrasound
An Interlaboratory Comparison
Background and Purpose Carotid ultrasound had modest accuracy in the North American Symptomatic Carotid Endarterectomy Trial (NASCET) of carotid endarterectomy in predicting severe carotid stenosis when a 250-cm/s peak systolic velocity (PSV) criterion was applied to different laboratories. We compared the performance of two independent laboratories using similar equipment (ATL-HDI Ultramark 9) but different interpretation criteria.
Methods Consecutive patients who underwent both color-coded duplex ultrasound and intra-arterial digital subtraction angiography were studied. PSV was determined with angle correction at the site of the tightest arterial narrowing. Carotid stenosis was measured on angiograms using the North American (N) method. Sensitivity, specificity, and positive (PPV) and negative (NPV) predictive values with 95% confidence intervals were calculated for each laboratory.
Results In 87 patients, 174 bifurcations were imaged. A 250-cm/s criterion was the best single predictor of a >70% N stenosis at one laboratory (sensitivity 93% [95% confidence interval, 85 to 101], specificity 86% [76 to 96], PPV 75% [62 to 87], and NPV 96% [90 to 102]) but had modest parameters at the other laboratory (50% [34 to 64], 87%, [77 to 97], 60 [44 to 76], and 91 [82 to 100], respectively). However, the diagnostic criteria routinely used in the second laboratory included different velocity values, which when applied decreased specificity by 17% but increased sensitivity by 35% (85% [74 to 96], 70% [56 to 84], 90% [81 to 99], and 77% [64 to 90], respectively).
Conclusions Despite the use of similar equipment, ultrasound grading of carotid stenosis is operator dependent and relies on different and individually validated criteria. Greater sensitivity of ultrasound screening is achieved by applying diagnostic criteria specific to each laboratory. Multicenter studies should use laboratory-specific criteria and a local validation process.
Although PSV is the most important component of the carotid Doppler examination,1 the grading of carotid stenosis with ultrasound should not be limited to this parameter only.2 3 As a screening test, carotid ultrasound should have an optimal tradeoff between sensitivity and specificity with the aim of identifying the highest percentage of patients with the potential of having a severe carotid stenosis. In NASCET, a 250-cm/s PSV criterion was applied to compare the ability of different ultrasound laboratories to predict severe carotid stenosis, and a high percentage of false-negative ultrasound studies was found.4
Several methodological simplifications compromised the validity of these observations3 ; however, the issue of compatibility of ultrasound studies from different centers deserves further evaluation. Besides the effect of collateralization and cardiac output, it is also erroneous to apply the same rigid velocity cutoff to different ultrasound machines, since the frequencies and geometry of the beams vary widely.5 Despite these differences, a positive correlation of ultrasound screening results is possible between laboratories,6 7 particularly if a prospective validation and standardization of diagnostic criteria is used.7 8
Some laboratories use similar equipment, however, with the only difference between them being the experience of vascular technologists and their practices of ultrasound examination. These differences theoretically may affect the choice and validation of the local criteria for grading carotid stenosis. Therefore, this study was undertaken to evaluate the performance of two independent but similarly equipped laboratories using a single velocity criterion and locally adopted combination criteria. Our hypothesis was that locally adopted criteria would increase the sensitivity of ultrasound screening to detect severe carotid artery stenosis and thus minimize false-negative study results.
Subjects and Methods
Consecutive patients who underwent both color-coded duplex ultrasound and intra-arterial digital subtraction angiography were studied prospectively. Both tests were interpreted independently from each other.
Ultrasound testing was performed at both laboratories with an ATL HDI Ultramark 9 equipped with the linear array L7-4 MHz transducer. The B-mode imaging frequency was 7 MHz, and the pulsed-wave Doppler frequency was 4 MHz. The angle-corrected velocity measurements were performed to obtain the stenotic jet velocity spectrum, as well as the velocity profile in the CCA, ECA, and ICA. At the second laboratory, diameter measurements of normal and stenosed vessels were also routinely performed on B-mode gray-scale images with superimposition of color-coded flow information.
At the first laboratory, a 250-cm/s PSV criterion was prospectively validated through the accreditation process and continuing quality assessment with sensitivity >90% and specificity >80% (D.S.B., unpublished data, 1994 to 1996). The following algorithm was used to diagnose >70% carotid stenosis. If ICA PSVs are ≥250 cm/s at the point of the tightest arterial narrowing, then the patient is diagnosed as having severe (>70%) stenosis. If the highest frequency shift corresponds to PSV <250 cm/s, and there is a clear view of the stenosed area with no shadowing in a normotensive patient, the degree of carotid narrowing is assessed as moderate. If there is an extensive shadowing artifact obscuring the view of the tightest narrowing, then distal flow characteristics are evaluated and compared with the proximal arterial segment, usually the CCA. ICA/CCA velocity ratios >3 are used to diagnose severe (>70%) stenosis in patients with low or high arterial pressure and in the presence of bilateral lesions. If tracing of peak systolic component is technically difficult, then a combination of other criteria is used (little or no diastolic component in CCA with longitudinal pulsation of the vessel; patients with low velocities <50 cm/s but extensive plaque on B-mode or the evidence of fresh thrombus [color flow gap] are identified as having 99% stenosis).
At the second laboratory, diagnostic criteria for severe carotid stenosis were also validated through a prospective quality control by comparing ultrasound data to invasive angiography (D.V., unpublished data, 1995 to 1996). The local criteria for the second laboratory include the following flow parameters. Severe (>70%) carotid artery stenosis is diagnosed if at least two of the following parameters are present: PSV >140 cm/s, end-diastolic velocity >125 cm/s, and ICA/CCA ratio >3, as well as >50% ICA diameter reduction on the transverse B-mode images. Decreased flow velocities <50 cm/s and an extensive lesion on B-mode are interpreted as near occlusion.
Intra-arterial digital subtraction angiography was performed within a month of ultrasound screening. Carotid stenosis was measured on a printed hard copy using the North American (N) method and expressed as a percent diameter reduction of the vessel.9 The diameter of the residual lumen was measured at the view with the tightest stenosis. The N denominator was determined at the first segment of the far-distal ICA with parallel walls beyond the poststenotic dilatation.
The accuracy parameters were calculated using a 2×2 table (screening test versus the gold standard) and included sensitivity and specificity, as well as PPV and NPV. The 95% confidence intervals were also calculated based on the sample size studied.
In 87 patients, 174 carotid bifurcations were imaged (laboratory 1, 40 patients; laboratory 2, 47 patients). At angiography, 31 vessels (17%) were normal, ie, had a diameter the same as or greater than the downstream diameter; 9 vessels (5%), including common and external carotids, were occluded, and 138 vessels (78%) had mild to severe stenoses.
A 250-cm/s criterion was the best single predictor of a >70% N stenosis at the first laboratory (sensitivity 93% [95% confidence interval, 85 to 101], specificity 86% [76 to 96], PPV 75% [62 to 87], and NPV 96% [90 to 102]) (Table⇓) but had modest validity and predictive value parameters at the other laboratory (sensitivity 50% [36 to 64], specificity 87%, [77 to 97], PPV 60% [44 to 76], and NPV 91% [82 to 100]).
However, the diagnostic criteria routinely used in the second laboratory include different velocity values to identify patients with >70% N stenosis. At this lab, severe carotid artery disease was diagnosed if at least two of the following parameters were present: PSV >140 cm/s, end-diastolic velocity >125 cm/s, and ICA/CCA ratio >3, as well as >50% ICA diameter reduction on the transverse B-mode images. When these locally adopted criteria were applied to the data set from the second lab, this method of ultrasound grading decreased specificity by 17% but improved sensitivity by 35% (85% [74 to 96], 70% [56 to 84], 90% [81 to 99], and 77% [64 to 90], respectively) (Table⇑), thus decreasing the number of false-negative studies. The first lab has previously validated10 and routinely uses the same 250-cm/s PSV criterion to detect >70% N stenosis, and therefore no further calculations were performed.
Our study showed that even if similar equipment is used, the rigid velocity criterion does not have the same validity and predictive values to grade carotid stenosis at different laboratories. The likely explanation is the difference in the sonographer’s technique and diagnostic algorithms used by the interpreters. Our hypothesis was that locally adopted criteria would increase sensitivity of ultrasound screening to detect severe carotid artery stenosis. Both laboratories previously validated their local diagnostic criteria, and when applied, these improved sensitivity by 35% (with specificity decreasing by 17%) at the second lab. This tradeoff is expected for a screening test, and the criteria should be set toward greater sensitivity to minimize false-negative studies. At the first laboratory, it was initially achieved with a 250-cm/s PSV criterion, while the second lab used a combination of lower velocity parameters to identify severe carotid stenosis.
Although 250 cm/s is often quoted as a criterion for >70% N stenosis,2 4 10 it may not be the best parameter for other laboratories. Hence, if the velocity cutoff is set too high, it would increase specificity at the price of sensitivity. For a screening test, the opposite tradeoff is desirable; therefore, a broader range of velocities should be included, such as in the University of Washington criteria and other combined criteria described in a recent review.9 The second lab in this study used a combination of PSV, end-diastolic velocity, and velocity ratios, as well as B-mode measurements, to interpret their studies.
We found differences in the performance of two laboratories equipped with the same color duplex scanners, and our data should prompt the laboratory or the interpreters to validate their own criteria. The choice of equipment cannot guarantee that the lab will produce good results with the criteria copied from another established lab. Validation of the diagnostic criteria is a part of the accreditation process developed by a multispecialty group to ensure proper training of personnel and assessment of the performance of vascular labs.8 Because grading carotid stenosis is an essential part of determining the risk of stroke, its accuracy should be assessed regularly with a prospective registry of ultrasound and other correlative methods, usually angiography.
Despite recent publications and overwhelming data on operator dependency of diagnostic criteria for ultrasound,2 3 the performance of the laboratories involved in the NASCET and Asymptomatic Carotid Atherosclerosis Study trials is being evaluated by retrospective application of the 250-cm/s criterion.11 Our study shows that even if similar equipment is used, this approach would artificially increase the number of false-negative studies. Therefore, multicenter assessment of the accuracy of ultrasound requires a different approach, with the emphasis on validation locally rather than application of uniform criteria across laboratories.7 Particular emphasis must be made to account for differences in local technique and diagnostic criteria.
A variety of criteria have been produced over the past decade for color duplex scanners.2 6 10 12 13 14 15 It may be difficult to decide which one to use as a template for establishing a new practice or in assessment of sonographer skills. One solution is offered by rigorous training at an established laboratory followed by an on-site prospective comparison of ultrasound findings with angiography. This self-assessment will allow modification of the criteria adopted from the established laboratory particular to the sonographer’s technique and angiographic verifications.
In conclusion, ultrasound grading of carotid stenosis is operator dependent and relies on different and individually validated criteria despite the use of similar equipment. A greater sensitivity (minimizing of false-negative studies) of ultrasound screening can be achieved by applying diagnostic criteria specific to each laboratory. Multicenter ultrasound data analysis should account for differences in the local diagnostic criteria used and utilize local validation to achieve the desired sensitivity rather than uniform criteria applied across centers.
Selected Abbreviations and Acronyms
|CCA||=||common carotid artery|
|ECA||=||external carotid artery|
|ICA||=||internal carotid artery|
|N||=||angiographic measurement according to the North American method|
|NASCET||=||North American Symptomatic Carotid Endarterectomy Trial|
|NPV||=||negative predictive value|
|PPV||=||positive predictive value|
|PSV||=||peak systolic velocity|
- Received January 2, 1997.
- Revision received March 12, 1997.
- Accepted March 12, 1997.
- Copyright © 1997 by American Heart Association
Spencer MP, Reid JM. Quantitation of carotid stenosis with continuous wave (C-W) Doppler ultrasound. Stroke. 1979;10:326-330.
Ringelstein EB. Skepticism toward carotid ultrasonography: a virtue, an attitude, or fanaticism? Stroke. 1995;26:1743-1746.
Eliasziw M, Rankin RN, Fox AJ, Haynes RB, Barnett HJM. Accuracy and prognostic consequences of ultrasonography in identifying severe carotid artery stenosis. Stroke. 1995;26:1747-1752.
Daigle RJ, Stavros AT, Lee RM. Overestimation of velocity and frequency values by multielement linear array Dopplers. J Vasc Tech. 1990;14:206-213.
Chang YJ, Golby AJ, Albers GW. Detection of carotid stenosis: from NASCET results to clinical practice. Stroke. 1995;26:1325-1328.
Howard G, Baker WH, Chambless LE, Howard VJ, Jones AM, Toole JF. An approach for the use of Doppler ultrasound as a screening tool for hemodynamically significant stenosis (despite heterogeneity of Doppler performance): a multicenter experience. Stroke. 1996;27:1951-1957.
Katanick SL. Accreditation of vascular ultrasound laboratories. In: Tegeler CH, Babikian VL, Gomez CR, eds. Neurosonology. St Louis, Mo: CV Mosby Co; 1996:484-488.
Nicolaides AN, Shifrin E, Dhanjil S, Griffin M. Duplex grading of internal carotid stenosis. In: Caplan LR, Shifrin EG, Nicolaides AN, Moore WS, eds. Cerebrovascular Ischemia: Investigation and Management. London, UK: Med Orion; 1996:101-110.
Alexandrov AV, Brodie DS, McLean A, Murphy J, Hamilton P, Burns PN. Correlation of peak systolic velocity and angiographic measurement of carotid stenosis revisited. Stroke. 1997;28:339-342.
Harbison J, Eliasziw M. Concern about safety of carotid angioplasty: response. Stroke. 1996;27:1436. Letter.
Strandness DE. Duplex Scanning in Vascular Disorders. New York, NY: Raven Press Publishers; 1990:92-120.
Neale ML, Chambers JL, Kelly AT, Connard S, Lawton MA, Roche J, Appleberg M. Reappraisal of duplex criteria to assess significant carotid stenosis with special reference to reports from the North American Symptomatic Carotid Endarterectomy Trial and the European Carotid Surgery Trial. J Vasc Surg. 1994;20:642-649.
Moneta GL, Edwards JM, Chitwood RW, Taylor LM, Lee RW, Cummings CA, Porter JM. Correlation of North American Symptomatic Carotid Endarterectomy Trial (NASCET) angiographic definition of 70% to 99% internal carotid artery stenosis with duplex scanning. J Vasc Surg. 1993;17:152-159.